Fingerprint image binarization based on multi-level block size

The effect of the self-adapt local threshold for fingerprint image is mainly determined by the estimate accuracy of the local threshold.For it is difficult to obtain the correct local threshold in the areas near singular points or in the ridge-valley regions with largely changed directions, a new method by multi-level blocks is proposed to get the local threshold of the fingerprint image,and then used to binarization of fingerprint image.The experimental results prove that this algorithm is steady,effective and robust,and the algorithm has remarkable results for low-quality fingerprint.